Microbial seedbanks and their evolutionary implications

Will Shoemaker

Background on dormancy

• Resources are finite

• Microbes respond by enterint a reversible state of low metabolic activity

• Ecological implications well-characterized(Jones & Lennon, 2010; Lennon & Jones, 2011)

• So are molecular & physiological mechanisms (Sturm & Dworkin, 2015)

from IPython.display import display, Image
display(Image(filename='Figs/DormRes.png'))

display(Image(filename='Figs/Migration_model2.png'))

• Formation of dormant pool = population stratification
• Impacts evolutionary inference
• Effect detectable on molecular level
• Similar to other demography-altering processes

display(Image(filename='Figs/CoalExample.jpg'))

display(Image(filename='Figs/DormCoal.png'))

Dormancy affects the geneology

• But how?

Dormancy and bean bag genetics

display(Image(filename='Figs/BeanBag.png'))

display(Image(filename='Figs/WFillustration.png'))

display(Image(filename='Figs/PopGenMut.png'))

WF-model without resampling

• Crude, simple, to the point
• divergence = $ N_{e} \mu * 1/N_{e} = \mu$
  • Assuming neutrality
  • $N_{e}*\mu$ mutations across the entire population
  • Each mutation has a $1/N_{e}$ chance of fixing.
  • Rate at which mutations fix is $\mu$ (Kimua 1968)

display(Image(filename='Figs/WF.png'))

Break the WF-model

display(Image(filename='Figs/WFdorm.png'))

Little data on microbial seed-banks & evolution

Recent surge of theoretical work

• Founded in coalescent theory (Blath et al., 2014; Kaj et al., 2001)

  • Wright-Fisher model with a seed-bank component

• Useful framework to examine neutral evolution of a microbial dormant population

  • $D_{FL}$
  • Normalized observed minus expected genetic variation

display(Image(filename='Figs/DFL1.png'))

display(Image(filename='Figs/DFL2.png'))

• Positive $D_{FL}$

• Naive view:

  • Selection maintains variation
  • Recent population bottleneck
  • Selective sweep

$$D_{FL} = \frac{S-a(n)\xi_{i} }{\sqrt{u_{n}S+v_{n}S^{2}}}$$

Beyond the coalescent...

• Start with population of all dormant indiduals
• #### Focus on core characteristics
  • No, or very slow, turnover
  • Long survivorship
  • Still has a mutation rate
    • No replication-associated mutation
    • Deconstructing the mutation rate

• Controversy on stress-induced mutation in dormant pops

• Seed-banks proposed as source of genetic variation in plants and microbes (Levin 1990; Sogin et al., 2006; González-Casanova, 2004)

• Dormancy considered trait that’s under selection (Dalling et al., 2011)

display(Image(filename='Figs/MAdist.png'))

$\mu$ and dormancy

• Old hyp: Environmental stress as selective force on $\mu$ (Foster 1998 Corzett et al. 2013)
  • Doesn’t change site-specific $\mu$
  • Doesn’t change beneficial $\mu$

• New hyp = Higher $\mu$ of dormant pops. as an obvious outcome
  • Mutation rate is a phenotype
  • Drift-barrier hypothesis (Sung et al., 2012)
    • Population genetics > biophysics
  • Limit on efficiency of selection (i.e. $\left | N_{e} * s \right | > 1.0$)
  • Difficult for selection to maintain unused pathways

Mutation-drift equilibrium

Expected neutral site heterozygosity

display(Image(filename='Figs/DriftMutEquilbJustAct.png'))

display(Image(filename='Figs/DriftMutEquilb1.png'))

display(Image(filename='Figs/DriftMutEquilb2.png'))

display(Image(filename='Figs/DriftMutEquilb3.png'))

But we don't infer $\theta$, we estimate it.

ATTTACGG
ATTTACAG     $\hat{\theta _{W}} = \frac{K}{a_{n}} = \frac{K}{\sum_{1}^{n-1}\frac{1}{i}} = \frac{3}{\frac{1}{1}+\frac{1}{2}+\frac{1}{3}+\frac{1}{4}} \approx 1.44$
ATCTACGG
ATTTACGG
ATTTTCGG

So we must infer $N_{e}$

• $N_{e} = \frac{\theta}{2\mu}$
• For $N_{e_{D}} > N_{e_{A}} = N_{e}$
  • Low $\mu$
    • Or high repair under stress
  • High $\theta$
    • Migration, MRCA pushed back
    • Recent bottleneck
• For $N_{e} = N_{e_{A}} > N_{e_{D}}$
  • High $\mu$
    • Population genetic limits to selection on repair fidelity
  • Low $\theta$
    • Recent population expansion

What does all this tell us?

• Simply having a portion of the population not reproducng and persisting (i.e. dormancy) should have an effect on the population's genetic makeup, with or without $\mu$
• Difficult, but not impossible to infer

What's next?

• Refine model
  • Two-way IBM model of migration
• Model selection
• Model effects of $\mu_{A}$ and $\mu_{D}$
• Model effects of different rates of migration between dormant and active subpopulations
• Compare to existing work
  • Refine model

$$\mathbf{Data} \leftrightarrow \mathbf{Theory}$$

Essential to tie in experimental work

• Core questions about basic population dynamics that haven't been answered
• Need for basic, controlled experiments to determine evolutionary impact of dormancy
• I'm not good enough at math to only do theory

Take flask picture

Demographic +/-'s with batch culture

:)

• If a seed bank forms, it stays there

:(

• Different rates of resuscitation/ dormancy (i.e. migration) during growth kinetics
• Have to track both population size and migration rate
  • Better to start with seed-bank as only demographic factor
• Risk loosing low-frequency alleles during transfers

Solution

Ensuring steady-state dynamics

• Chemostats
• Adjust dilution rate to allow for a constant total population size and different dormant sub-population sizes across treatments
• Problem: I don't know anything about chemostats

Monitoring subpopulation sizes

• Flow cytometry
  • DNA associated dye
  • Respiration associated dye
  • Membrane permeabilization associated dye

display(Image(filename='Figs/FlowExample.jpg'))

What organism?

guess...

display(Image(filename='Figs/IMG_0530.jpg'))

Janthinobacterium sp. KBS0711

• Exhibits population dynamics associated with organisms capable of entering long-term dormancy
  • Does so in the wild (Alonso-Sáez etal., 2014)
• Genes associated with the ability to contend with a stressful state
  • PhoR-PhoB
• MA currently underway
• Awesome pub (Shoemaker et al., 2015)